A Strong Law for the Largest Nearest-Neighbour Link between Random Points

From MaRDI portal





Let \(X_1,X_2,X_3,\dots\) be independent random points in \(\mathbb{R}^d\) with common density function \(f\). Let \(R^n_{t,k}\) denote the distance from \(X_t\) to its \(k\)th nearest neighbour among the first \(n\) points, and let \(M^d_{n,k}= \max_{i\leq n} R^n_{t,k}\). The author considers a general class of densities, bounded away from zero on compact support with smooth boundary. He obtains a strong law for \(M^d_{n,k}\) as \(n\to\infty\), \[ n\theta M^d_{n,k}/\log n\to \max\Biggl(\Biggl(\min_\Omega f\Biggr)^{-1},\;2(1- 1/d)\Biggl( \min_{\partial\Omega} f\Biggr)^{-1}\Biggr)\quad\text{a.s.}, \] where \(\theta\) denotes the volume of the unit ball and \(\partial\Omega\) the smooth boundary of \(\Omega\). If instead the points lie in a compact smooth \(d\)-dimensional Riemannian manifold \(K\), then \[ n\theta M^d_{n,k}/\log n\to \Biggl(\min_k f\Biggr)^{-1}\quad\text{a.s.}. \]




Cited in
(24)








This page was built for publication: A Strong Law for the Largest Nearest-Neighbour Link between Random Points

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4487496)